The second is the test results retained too many valid digits. The first reason is that they do not understand the rules for calculating the number of significant digits, unintentionally retained, the second reason is deliberately retained in the hope that this "improve" the accuracy of the results.
Third, in the external inspection report or submitted to the competent authorities at all levels of the summary of the original data reproduced in the test,? I do not know the test method requires reasonable retention of the number of valid digits of the test results, but also do not know the judgment (limited value, index value) standard conversion unit of measure, or conversion unit of measure when the arbitrary increase or decrease in the number of valid digits, these practices will affect the impartiality of the testing organization, authority.
Therefore, a key link in the quality management of the quality inspection agency is the quality control of test results, it is to ensure the accuracy of the test data, test conclusions of the scientific and impartiality, and has the important link of traceability.
Quality inspection agencies on the quality control of test results of the technical points I think should consider the following aspects:?
1 strict data processing and control?
The application of computers in scientific research, experiments and all aspects of management has become a development trend, the inspection agency, the inspector in addition to the daily inspection work, a considerable part of the time spent on the management of equipment and materials. In addition, in the test work, has been widely utilized microcomputer data acquisition, results processing and test report output. How to use computers for effective scientific management? This requires inspectors to have certain management knowledge and experience, as well as mastering the practical application of computer operations, with the help of computers to standardize the management, scientific, improve efficiency. Laboratory should have appropriate calculations and data conversion and processing regulations and effective implementation.
2 Correct recording of measurement observations?
In practice, many inspectors due to the concept of fuzzy and unclear how to accurately and accurately record the measurement of the observed value, thus affecting the accuracy of the final test results, the correct recording of the measurement of the observed value of the principles to be grasped is that: first of all, to correctly understand the concept of the effective number, which refers to the measurement of the number of the actual measurement of the number of the value of the measurement of all the accurate value and a doubtful value. Secondly, the number of effective digits to record the results of measurement should be consistent with all measuring instruments, instrumentation, measurement accuracy, can not be arbitrarily taken more or less, the accuracy of the instrument here also includes the effective value of the standard substance, the following analysis of several types of commonly used instruments, gauges, standards, as an example.
(1) weighing with an analytical balance (minimum divisional value of 0.1mg), the effective number of digits can be recorded to the fourth decimal point, such as 2.1453g at this time the effective number of digits for the five; weighing 0.5687g, then for the four, with the hundredths of a day (minimum divisional value of 0.01g) weighing 25 grams of specimen, should be recorded as 25.00 g, recorded as 25g! is just wrong.?
(2) constant buret and pipette recorded to milliliters as units of 2 digits after the decimal point; 2ml or less of microburet, its readings should be recorded to milliliters as units of 3 digits after the decimal point. That is to say, the buret can take up to 4 valid digits, such as 10.23 ml, sometimes only 3 digits, such as 5.23 ml, and sometimes there are 2 digits and 1 digit, such as 0.48 ml and 0.03 ml, etc. 100~1000 ml volumetric flasks should be recorded to 1 digit after the decimal point, and volumetric flasks of less than 50 ml should be recorded to 2 digits after the decimal point. For example, a single labeled line A 50 ml volumetric flask, the exact volume of 50.00 ml, the effective number of digits for four. Colorimetric tube in the test of dilution to scale operation can be regarded as a volumetric flask capacity, can be taken to 4 digits, but it should be noted that its accuracy is not as good as volumetric flasks.
(3) spectrophotometer minimum division value of 0.005, therefore, the absorbance can generally be recorded to the third decimal place, the effective number of digits is generally up to 3.?
(4) with a computer processing system of analytical instruments, often according to the computer's own settings to print or display the results, there can be many digits, but this does not increase the precision of the instrument and the number of readable effective digits, in a series of operations, the use of a variety of measuring instruments, the effective number of digits to the least number of a recording instrument said. Therefore, the chromatography of the general take 3 effective digits, up to 4, such as liquid-phase UV detector is actually a spectrophotometer, gas-phase such as FID detector, in fact, is the current detector, although the instrument gives the signal value of many digits, but the effective number of the same as the general ammeter? At the same time the effective number of chromatography is limited by the effective number of bits of the injection needle, such as 1.00 microliters for the gas phase and 20.00 microliters for the liquid phase.?
(5) buy the standard solution is generally 4 effective number, we are diluted, especially after high dilution, generally to reduce the effective number of reasonable. Such as their own preparation of the standard solution, but also pay attention to the content of the original reagent with the effective number of digits, such as its labeled as ≥ 99.95%, can take 4, such as ≥ 99.9%, can only take 3.?
3 Accurate calculation of test results?
The inspector in the test not only to accurately determine a variety of data, the correct record, but also according to the rules of arithmetic to accurately calculate the test results. Because the value of the test results not only indicates the content of the item being measured how much, but also reflects the detection method, the accuracy of the inspection process, so it is important to correctly handle the test data. First, the testing staff of the detection method should be correctly understood in the calculation formula to ensure that the calculation of test data and units of measurement between the conversion without error, the results of the calculation of self-calibration and review. Second, the effective number of test results should be consistent with the provisions of the test method? Calculation of the middle of the data obtained from the effective number of bits should be retained one more.
Specific operational procedures are as follows:
(1) First of all, according to the testing process of the analytical method and the accuracy of the instrument precision, to determine the effective number of digits involved in the operation of the value of the first operation, according to the "GB8170-2008 numerical modification of the rules and limits of the value of the representation and determination of" modification to arrive at the original test data the process to grasp the confirmation of the effective digit
This process is to grasp the confirmation of the effective number, the number of revision and the principle of effective number of operations.
Acknowledgement of valid numbersA. In determining the number of valid numbers, whether the number "0" is a valid number depends on where it is located in the data. In front of the decimal point "0" only plays a role in positioning, not a valid number, the middle of the data and the last "0" is a valid number. b. In analytical chemistry, often encountered pH, pM, pC and other logarithmic values, these logarithmic values of the number of valid digits only depends on the number of digits after the decimal point, and the integer part has nothing to do with the integer part of the integer only plays a role in positioning, not a valid number. For example, pH 4.70, the effective digit digits of the logarithmic mantissa of the digit after the decimal point, because the logarithmic mantissa of the integer is only related to the true 10th power. c. In the process of calculation, some non-deterministic values will be encountered, such as dilute multiples, concentrated multiples, fractions, etc., whose effective digit digit digits can be regarded as infinite multi-digit. d. For the first digit of a number greater than or equal to 8, the effective digit digit digit digit digit can be counted one more than the actual number of digits of the data. Principles of Numerical Modification "GB/T 8170-2008 Rules for Numerical Modification and Representation and Determination of Limit Values" stipulates the rules for modification of numerical values, the method for representation and determination of numerical limit values, and the method for comparing the measured or calculated values with the limit values specified in the standard. After the determination of the effective number of digits, the excess number of digits for revision, the principle is rounded up to six, when five odd into even rounding. Revision should pay attention to when five, such as five behind the zero into one, five behind the zero, five of the first odd into one, even or zero is about to go, revision, but also can not be continuous revision. The principle of effective numerical operations effective numerical operations can only retain l uncertainty of the number of digits. Addition and subtraction operations, the number of valid digits retained in the results of calculations to the least number of decimal places in the data, i.e., the absolute error of the largest prevail. When multiplying and dividing, the number of significant digits retained? The number of valid digits retained is determined by the number of digits with the least number of decimal places in the data, i.e., the one with the largest relative error?(2) Then according to the detection limit of the test standard, to determine the calculation of the original test data should be retained in the last digit, for example, a method of the minimum detection of quality concentration of 0.02mg / L, the analytical results reported 0.088mg / L is unreasonable, should be reported as 0.09mg / L.?
(3) Finally, according to the requirements of the test standard for the number of bits retained in the test results, the data for selective revision, such as its bits have been in line with or less than the number of method provisions, then remain unchanged, otherwise, and then revised, a one-time revision to meet the requirements.?
4 Reasonable reporting and determination of test results?
(1) Examples of test results report?
Constant component analysis, the content of ≥ 10% of the results expressed in four digits; content of 1 ~ 10% expressed in three digits, the content of ≤ 1% micro-component analysis is usually expressed in two to three digits of the analytical results. If the titration method or mass method test, when the concentration of the measured substance content is high and the sampling volume is large, the relative error of measurement can be as low as one thousandth, then the test results can be reported as 4 digits. The use of various types of instrumental analytical testing, the number of valid digits of the test result is generally 2 to 3, in the vicinity of the detection limit is often 1, so the laboratory reviewer should pay attention to the test report should not appear in the test report of 5 or more valid digits of the test results.
The results of the unit of measurement should be used in the Chinese people*** and the country's legal unit of measurement, and generally require and determination of standards, such as product standards or limit standards, to maintain consistency, in order to compare and judge, such as the method of standard GB/T5009.22 in the 9 test, rice Aspergillus flavus B1 for 8.6ng / g, the test report should be limited to the standard GB2715-2005 The provisions of the conversion report for 8.6ug/kg; such as the method standard GB/T5009.15-2003 the first method of testing, the cadmium content of wheat is 35 ug/kg, the test report should be in accordance with the provisions of the limited standard GB2715-2005 conversion report for 0.035mg/kg;?
Analytical results below the limit of detection, can be expressed as "not detected", and indicate the limit of detection value or the lowest limit of detection to report the results, such as 0.02mg/kg.?
Parallel sample results within the allowable deviation, the report with its average value of the results of the measurement, and report the results of calculations expressed to the number of decimal places or significant digits, first, the results should be retained to the detection method requires significant digits; second, the number of significant digits of the measured value should be able to meet the requirements of health standards. Such as soybean, rapeseed and other crude oil solvent residue GB2716-2005 health standard requirements is ≤ 100mg / kg, while the test method GB/T5009.37 requirements to retain three valid digits, such as making the value of 100.3 mg / kg, can be reported as 100mg / kg. Conclusion of the qualified, and soybean, rapeseed and other three, four grade oil solvent residue health standard requirements is ≤ 50mg / kg, the number of valid digits of the measured value should be able to meet the requirements of health standards. The requirement is ≤50mg/kg, if the value is 50.34 mg/kg, it can only be reported as 50.3mg/kg, the conclusion is unqualified.
When the concentration of the measured substance is too high or too low, should use the scientific method of counting, because the effective number of digits from the measuring instrument, reflecting the degree of accuracy of the measurement instrument, so the unit change should not change the number of digits of the effective number of digits, such as the test data 12.2g/kg if the standard to be used in mg / kg and ug / kg as a unit, although from the point of view of the mathematical point of view, it can be recorded as 12200mg / kg and 12200000ug / kg, but from the measurement point of view, this practice changes the number of significant digits, it is wrong to use scientific notation can be guaranteed in the unit of change, the number of significant digits of the unchanged, so the test report requires the use of scientific notation as far as possible to express the experimental data.
Usually, the experiment to do double experiments, parallel samples of the results within the allowable deviation, the report with the average of the results, but the test analysis there are some exceptions, such as the color of grease, flour and other magnetic metal substances to be determined by the high results of the test results of the results of the results of the test, rather than the average, which should be noted, do not deal with the data according to the customary, otherwise it will also lead to The test conclusion is wrong.
(2) the method of determining the test results of the national standard "GB/T 8170-2008 numerical modification rules and limit values of the representation and determination" points out that in determining whether the measured value and calculated value in line with the standard requirements, there are two methods of discrimination, that is, the modification of the value of the comparison method and the full value of the comparison method. The former is to determine the value or calculated value after modification of the value and the standard limit value comparison, modification of the number of digits and the standard limit number of the same, the latter is to determine the value or calculate the value without modification and the value of all the figures and the standard limit value comparison, as long as the limit of the number of figures beyond the prescribed range, regardless of the extent of the size of the exceeding, are judged to be not in line with the standard requirements.
GB/T 8170 stipulates that "the standard or related documents, a variety of limit values, including values with limit deviation value without special provisions, should use the full value of the comparison method, such as the provisions of the revised value of the comparison method, it should be explained in the standard." Such as GB/T5009.15 test cadmium, measured residue content of 0.24mg/kg, which exceeds the GB2715 national standard of 0.2mg/kg of the numerical limits of the detection method of the effective number of provisions for the two, should be used to the full value of the comparative method, so it can not be trimmed to 0.2mg/kg, the results should be judged as unqualified. At present, in our national standard, it is difficult to see the explicit requirement to use the modified value comparison method? So we can only use the full value comparison method in our daily judgment.
(3) measurement uncertainty assessment due to the measured or calculated value of its own is not 100% accurate and reliable, it is impossible to avoid from the measurement equipment, environment, personnel, measurement methods and the object to be measured and other uncertainties. Daily use of the full value of the comparison method is likely to cause unreasonable or unfair results for the party being examined, easy to lead to a lot of controversy, and even some judicial disputes, therefore, China's National Laboratory Accreditation Board in the "assessment and reporting compliance rules" stipulates that "when the test is designed to compare the results with the qualifying value, rather than measuring a specific value, the uncertainty must be assessed. Uncertainty must be assessed." Meanwhile, in GB/T27025, General Requirements for the Competence of Testing and Calibration Laboratories, it is also stipulated that "when the uncertainty is related to the validity or application of the test results, or when the uncertainty affects compliance with the limits of the specification when the test report also needs to include information about the uncertainty." ?
5 Summary ?
Now the use of calculators has been popularized in the calculation process to take a few more digits and not much effort, the calculation will not be difficult, but the correct expression of the results of the experiment is still worthy of attention, the inspector should be able to accurately determine the experiments have a few valid numbers, the correct results of how to express, the management of the reviewer or the technical and technical responsible person, responsible for the quality of the person responsible for the review of the gatekeeper of each link of the test data Whether to meet the requirements.Because as a third-party testing organization for external services, the rigor of the test report, the scientific focus on the accuracy of the test data and test conclusions, to ensure that the scientific and accuracy of the test data, quality inspection agency managers should take a variety of ways to focus on cultivating and improving the quality of professional and technical personnel and management personnel, training to highlight the importance of basic skills and high-precision skills, the work in accordance with national standards and laboratory quality control. Work in accordance with national standards and laboratory quality control specifications for operation, mastering the various aspects of testing data processing and quality control of technical points in order to improve the level of testing technology and laboratory management, better to the community and government departments to issue real test results and accurate test conclusions. Answers from